UV-visible absorption cross sections of nitrous acid

Transcription

1 JOURNAL OF GEOPHYSCAL RESEARCH, VOL. 105, NO. Dll, PAGES 14,585-14,592, JUNE 16, 2000 UV-visible absorption cross sections of nitrous acid J. Stutz, 1 E.S. Kim, and U. Platt nstitut fiir Umweltphysik, Heidelberg, Germany P. Bruno, C. Perrino, and A. Febo CNR nstituto nquinamento Atmosferico, Rome, taly Abstract. Nitrous acid, HONO, is a source of OH radicals in the polluted atmosphere. Although the atmospheric chemistry of HONO is qualitatively understood, not much quantitative information exists. The magnitude of the OH production by HONO photolysis depends on the spectrum of its absorption cross sections; therefore the knowledge of Ch ono0 ) is essential. The spectrum of the differential cross sections C ' ONO0 ) is needed to detect HONO in the atmosphere by differential optical absorption spectroscopy (DOAS). Here we present measurements of the HONO UV-visible absorption cross sections with a spectral resolution better than 0.1 nm and a high signal-to-noise ratio. The maximum value of the absorption crossections is Ch ono(354 nm)- ( ) x cm: and agrees well with literature data. Nevertheless, calculations based on data from this work and on literature data reveal that an uncertainty of- 15% remains for the HONO photolysis rates. The new C } ONO0 ) has been employed in DOAS measurements in Milan, taly. 1. ntroduction The possible role of nitrous acid, HONO, in the atmosphere has been discussed since the mid-1970s [Johnston and Graham, 1974; Cox, 1974; Nash, 1974] and more intensely after its positive identification in the atmosphere in 1979 by Perner and Platt [1979]. The importance of HONO in the polluted atmosphere is based on its photolysis to NO and OH radicals: (R1) HONO + hv - OH + NO. During the first hours of the morning and during wintertime periods, OH production by ozone and formaldehyde photolysis is small. The HONO mixing ratio, in contrast, reaches its maximum during the last hours of the night, and OH production through its photolysis in the morning is strong enough to trigger and support OH-catalyzed oxidation processes during these periods [Harris et al., 1982]. Since there is no reason to assume that the chemical reactions producing HONO should not proceed in daylight, it is suspected that HONO photolysis is also an important source of OH radicals during the day (for recent reviews of HONO chemistry see Lainreel and Cape [1996], and Febo and Perrino, [1998, and references therein]). Nighttime mixing ratios of HONO in the urban atmosphere of up to -10 ppb have been reported [Lainreel and Cape, 1996]. Daytime HONO observations are sparse, presumably because of the much lower concentration. Today it is believed that HONO is mainly formed heterogeneously with a rate that Now at Department of Atmospheric Sciences, University of California, Los Angeles. Copyright 2000 by the American Geophysical Union. Paper number 2000JD /00/2000JD $ ,585 appears to be first order in NO2 and H20 [Svensson et al., 1987; Jenkin et al., 1988; Harrison et al., 1996]. The stochiometry can be summarized to (R2) NO2 + NO2 + H20 > HONO + HNO3. Studies aimed at determining the exact nature of this mechanism and the possible influence of the chemical characteristics of the surfaces on which it takes place are in progress. Many aspects of the atmospheric chemistry of HONO remain unclear, including details about the formation and re- moval mechanisms and a quantitative assessment of its impact on the oxidative properties of the atmosphere. Also the spectrum of the absorption cross sections, the most important parameter for the quantification of the OH production by HONO photolysis, still has a considerable uncertainty. This uncertainty also determines the error of differential optical absorption spectroscopy (DOAS) measurements [Platt and Perner, 1983; Platt, 1994], one of the most successful techniques to measure HONO in the atmosphere. Several methods to detect atmospheric HONO are currently used: Among the systems based on ion chromatographic analysis of NO2-, annular dry denuders are most widely employed [Allegrini et al., 1987; Appel et al., 1990; Febo et al., 1993, 1996]. Another method, the wetted wall denuders, has been presented in the last years [Vecera and Dasgupta, 1991; Simon and Dasgupta, 1995]. A continuous method based on the conversion of nitrite to NO followed by a detection by chemoluminescence has also been used [Kanda and Taira, 1990]. These methods are user-friendly and relatively inexpensive, but special care has to be taken to avoid cross sensi- tivities with other atmospheric species; therefore they require much experimental work [Febo et al., 1993]. A high time resolution, selectivity, and sensitivity characterize spectroscopic methods in the UV-visible and R. The most successful of these techniques is the DOAS [Platt and Perner, 1983; Appel et al., 1990; Platt, 1994], which is based on UV-visible absorption spectroscopy in the open atmosphere. Light path

2 14,586 STUTZ ET AL.: UV-VSBLE CROSS SECTONS OF NTROUS ACD lengths of up to several kilometers are used to reach the detection limits of the order of some tens of parts per trillion. The principle of DOAS is the identification and quantification of trace gas concentrations by analyzing the narrowband absorption structures, avoiding interferences with broadband extinction processes like Rayleigh and Mie scattering. DOAS requires a numerical analysis of the atmospheric absorption spectra, which can contain superimposed absorptions of several other trace gases, by a least squares fit of reference spectra of the pure trace gases [Stutz and Platt, 1996]. The performance of this fit depends on the agreement of the spectroscopic structures in the atmospheric spectrum and the respective trace gas reference spectra. For gases like HONO, which are experimentally difficult to produce, the reference spectrum is simulated numerically, based on a literature absorption cross-section spectrum and the instrument function of the field experiment [Platt, 1994; Stutz and Platt, 1996]; the latter can be determined by measuring atomic emission lines. The most important spectroscopic requirements of an absorption cross-section spectrum used for this simulation are a high spectral resolution or at least the knowledge of the instrument function of the system recording the cross-section spectrum and a high accuracy in the spectral position of the absorption bands. Especially for HONO, these properties are essential due to possible spectroscopic interferences with the always present NO2 absorption in atmospheric spectra. Any inaccuracy in the HONO absorption cross-section spectrum directly translates into a systematic error of the measured atmospheric concentrations. Several absorption cross-section spectra for HONO have been published [Johnston and Graham, 1974; Cox, 1974; Cox and Derwent, 1976; Stockwell and Calvert, 1978; Perner and Platt, 1979; Vasudev, 1990; Bongartz et al., 1991, 1994; Febo et al., 1996; Pagsberg et al., 1997; Brust et al., 1999]. They all differ considerably, and many are of poor or unknown spectroscopic quality, with spectral resolutions ranging from 0.1 nm to 1 nm. The main difficulty in measuring the cross sections is the production of NO2-free HONO. Some of the literature data [Johnston and Graham, 1974; Stockwell and Calvert, 1978] were obtained by observing HONO in an NO2-water vapor mixture, while others [Cox and Derwent, 1976; Bongartz et al., 1991, 1994] used the reaction of sulfuric acid with sodium nitrite to produce HONO. Both approaches face two major problems. (1) t is difficult to determine HONO concentration in the reactor by analytical methods due to possible interferences with other nitrogen compounds. (2) The NO2 absorptions overlying the HONO bands have to be corrected in the spectrum. Also the temporal stability of the HONO concentration is a problem. The need to record the whole spectrum in a short time period often results in poor spectral resolution and signal-to-noise ratio. The recent development of an apparatus to produce highly pure and constant HONO flows in one of our laboratories [Febo et al., 1995] made it possible to overcome most of these problems. We present here measurements of the spectrum of the HONO absorption cross sections in the UVvisible spectral range ( nm) with a high spectral resolution and known instrument function. The influence of this new cross-section spectrum on DOAS measurements and atmospheric photochemistry of HONO will be discussed. HC1, + H20 Thermostatic Stirrer Lamp Bath c NaNO 2,, Spectrograph ic Solution Quartz Cell NO Monitor Figure 1. Schematic diagram of the experimental setup to measure the HONO absorption cross sections. 2. Experimental Methods A constant flow of 0.5 L min ' at atmospheric pressure of high-purity HONO in humidified nitrogen was generated by a system (Figure 1) described by Febo et al. [1995]. t is based on the reaction of a gaseous HC1 and solid sodium nitrite: (R3) HCl(g) + NaNO2(s) - HONO(g) + NaCl(s) A humidified flow of high-purity nitrogen is passed through a Teflon tube (outer diameter, 2 mm; wall thickness, 0.2 mm; length, 225 cm) which is coiled in a 1 L gas-tight vessel filled with an aqueous HC1 solution. The vessel is placed in a thermostatic bath with +_ 0.1 øc temperature control. Diffusion of HC1 through the Teflon tube transports small amounts of HC1 into the N2- H20 gas stream. The amount of HC1, and therefore also the HONO production, can be adjusted by changing the length of the Teflon tube, the HC1 concentration in the solution, and the temperature. The N2- HC1- H20 gas mixture enters a glass cell thermostated at 50øC and containing about 1 g NaNO2 powder, which is continuously mixed by a magnetic stirrer. The system has been extensively tested before, and shown to give stable flows of constant HONO concentrations. As stated by Febo et al. [1995], the stability in earlier experiments was better than 0.2% over a period of several hours, and the purity was better than 99%. Only at HONO concentrations in the range of 10 4molecules cm -3 does the decomposition of HONO, which is suspected to pro- ceed via (R4) HONO(g) + HONO(ads) - NO + NO2 + H20 introduce small impurities of NO and NO2. The concentration of HONO in the absorption cell was determined by two methods: (1) During the spectroscopic measurements a modified gas phase NO/NOy chemiluminescence analyzer, described Ly Febo and Perrino [1991], was operated to observe the stability of the HONO concentration. Since the purity of the gas mixture was better than 99%, this also gave a first indication of the HONO concentration in the absorption cell. (2) The exact HONO concentration was determined by bubbling the gas stream through a slightly basic solution, leading to a complete dissociation into H + and NO2-. The solution was then analyzed by means of a Dionex

3 _ STUTZ ET AL.' UV-VSBLE CROSS SECTONS OF NTROUS ACD 14,587 Table 1. HONO Concentrations Before and After the Cell Determined by on Chromatot raphy Run [HONO], 1014 molecules cm '3 [NO2], 10 TM molecules cm '3 Before Cell After Cell Average The NO2 impurities were determined by absorption spectroscopy. All errors in this table and the rest of this paper are twice the standard deviation. DX500 ion chromatograph. To determine the loss of HONO on the wall of the cell due to self-reactions, the concentration was determined before entering and after leaving the absorption quartz cell. Table 1 gives an overview of the values determined for the individual runs. Possible NO2 impurities in the gas stream were observed spectroscopically by turning the grating of the spectrometer to 440 nm where the strongest differential NO2 absorptions are found. The results of these measurements are also shown in Table 1. The optical setup (Figure 1) consisted of a standard halogen lamp collimated by two f= 150 mm quartz lenses. The light beam was fed through a cm long quartz cell with 5 cm diameter and focused onto the 80 gm wide entrance slit of a Czerny-Turner type f= 0.5 m spectrograph (Acton Spectra Pro 500, with a 2400 g mm - grating) by an- other f= 150 mm lens. A photodiode array detector (Hoffmann Messtechnik) based on a Hamamatsu 1024-diode array (Type SF N) was used to record a spectral range of -17 nm. To determine the spectrum of the HONO absorption cross sections from 290 nm to 420 nm, 11 individual spectral intervals were recorded. These spectra overlapped by-5 nm. Typical integration times for individual scans were between 0.3 and 1.2 s. A total of 50 scans were added per spectrum to reach a better signal-to-noise ratio. Special care was taken to determine the precise spectral position of the HONO absorption spectrum. Spectra of Hg and Ne atomic emission lamps and a quartz cell filled with high-purity NO2 in oxygen placed in the parallel light beam were recorded. The dispersion varied slightly in the different wavelength windows, as was expected by the optical setup of the spectrograph. A dispersion of nm pixel - ( nm mm - ) was found in the UV (-290 nm), decreasing to nm pixel - (0.062 nm mm - ) in the blue wavelength region (-420 nm). The dispersion was found to be slightly nonlinear; therefore a quadratic function was used to describe it nm, where the error describes the variation over the whole wavelength range. The instrument function is shown for the example of the 334 nm Hg line in the inset of Figure 2. The asymmetrical shape is due to the optical properties of the spectrograph. The overall shape of the HONO absorption cross-section spectrum is not influenced noticeably by the line shape since the bands are much broader than the mercury line width. The whole spectrum of the absorption cross sections was assembled from the 11 overlapping wavelength intervals. The main problem when assembling the spectrum was found to be the stability of the lamp intensity during a measurement cycle of around 1 hour length. We determined the stability to be better than 0.2%. Although this variability is quite small, it can be detected when comparing intensities in two different wavelength intervals. We therefore had to correct these variations by multiplying the intensities in the wavelength intervals by factors of The intensity differences at the joining points of the individual spectra after the correction - lx10 ' ' The accuracy of the center wavelength of an individual wavelength (nm) 17 nm spectral interval was determined conservatively to Figure 2. Spectrum of the absorption cross section of HONO 0.02 nm or -1 pixel (all errors in this paper are considered recorded at a concentration of 11.1 x 10 4moleculescm -3. twice standard deviation). The spectra of the two emission The residual NO2 in the gas stream was less than 0.1% lamps were also used to determine the instrument function [NO2]/[HONO] and was subtracted from the spectrum. The and spectral resolution. Since the resolution is dominated by inset of the figure shows the Hg emission line at 334 nm the slit width, it changes slightly with varying dispersion. The measured by the spectrograph. The vertical lines indicate the resolution averaged over 40 Ne and Hg emission lines was joining wavelength of two spectral regions. v E 5xl 0 ' 9 4xl 0 ' 9 3xl 0 t 2x10 ' " 0 ' i, i

4 14,588 STUTZ ET AL.' UV-VSBLE CROSS SECTONS OF NTROUS ACD were indistinguishable from the noise in the spectra. To as- semble the spectrum, we also assumed (405 nm)= 0 cm 2. A small absorption can not be excluded at this wavelength and has to be considered as a possible baseline error of the absorption cross-section spectrum. Figure 2 shows the spectrum of the absorption cross sections measured at 11.1 x 10 TM molecules cm -3 HONO in the cell. Vertical lines indicate the respective joining wavelengths of the individual spectral windows. Small amounts of NO2 were observed at higher HONO concentrations. Although the absorptions were 2 orders of magnitude smaller than those of HONO, we used measurements of a NO2 reference cell to remove the differential and absolute absorption of NO2. ' ' ' ' ' '. a , nrn/ ' ' ' ' ' ' 3. Results 3.1. Spectral nformation The spectroscopic quality of our absorption cross-section..,o 4 spectrum was confirmed by comparing the spectral positions of the most prominent absorption bands (Table 2) with the rn two literature data sets of the highest resolution [King and Moule, 1962; Bongartz et al., 1991]. While Bongartz et al. [1991] used a spectral resolution of MA3,-3500, similar to ours, the resolution of King and Moule [1962] was about 10 '"' 2 times higher. The errors of our values were determined by the 0 uncertainty of the dispersion (see above) and the error in the determination of the spectral position of the absorption band maximum, which was estimated to be < pixel ( nm). HONO concentration [1014cm'3] n general, there is an excellent agreement within the error Figure 3. Examples of the linear dependence of the (a) differlimits for the strongest HONO bands. Our values are closer to ential and (b) absolute absorption of the most prominent abthose of King and Moule [1962] than to the results of Bon- sorption bands from the HONO concentration in the cell. The error bars in Figure 3b indicate the error in the intensity of the lamp. The errors in Figure 3a are too small to be shown. The results of the linear fit for the differential cross section are Table 2. Comparison of the Spectral Position of the Different '(354.2 nm, nm)= ( ) x cm 2, HONO Bands in Vacuum Wave Numbers of the Cis and 0'(354.2 nm, nm)= ( ) x cm 2, and Trans somers 0'( nm, nm)= ( ) x 10 -]9 cm 2. The Trans Bongartz et al. King and Moule [1991] [1962] This Work 0 26,035 26, , ,153 27, , ,220 28, , ,249 29, , ,227 30, , ,158 31, , ,076 32, , Cis 0 26, ,430 27, , ,496 28, , ,513 29, , ,502 30, , ,456 31, , ,344 32, , Values are in units of cm '. The values of Cox and Derwent [ 1976] also agree well with those listed in the table, but were omitted due to their low spectral resolution. results of the absolute absorption cross section in Figure 3b are 0(342 nm)= ( ) x cm 2, (354 nm)= ( ) x 10 -]9 cm 2 and 0(368 nm)= ( ) x cm 2. n all cases the offset was statistically insignificant. gartz et al. [1991]. The spectral position of bands falling in two spectral intervals in our measurements agrees within the error limits shown in Table 2, proving the quality of the determination of the dispersion and the validity of our approach of assembling a single absorption cross-section spectrum from the individual spectra Differential Absorption Cross Sections There is no unique definition of the differential absorption cross section c ' 0 ), since it depends on the filtering procedure separating broad from narrow (differential) absorption structures. A closer analysis of c '(3,) is nevertheless important since it is the parameter needed for DOAS measurements and it is independent of uncertainties in lamp intensity and other spectroscopic problems. We have chosen the definition of the differential absorption cross section according to [Bongartz et al., 1994]: 6' 0 ],3,2) = 6(3,1)-6(3,2).

5 STUTZ ET AL.' UV-VSBLE CROSS SECTONS OF NTROUS ACD 14,589 Table 3. Comparison of the Differential and Absolute Absorption Cross Sections with Literature Data Spectral ( at 354 nm, ( at 354 nm Difference Difference From Resolution, 10 cm 2 With a Resolution From This the ( ' of This nm of 1 nm, 10 cm 2 Work, % Work, %c Johnston and Graham [1974] Cox and Derwent [ 1976] < Stockwell and Calvert < [1978] Perner and Platt [ 1979] a 13 Vasudev [ 1990] (b) Bongartz et al _+ 0.1 [1991, 1994] Febo et al. [ 1996]...b...a 8.2 Pagsberg et al. [1997] _ _+ 0.5 Brust et al. [ 1999] _ _+ 5 d This work _ _ a Only cy' was given in this publication. b Unknown. c The errors reflect the uncertainty of the fit. d Value calculated with the netho described by Bongartz et al. [1991]. Figure 3a shows the differential optical density ln(l'/h)')/l (with '=(X1)/(X2), /0'=/0(X1)//0(X2) and L the length of the cell) of the three strongest HONO absorption bands as a function of the HONO concentration in the cell. The curves show an excellent linear dependence on HONO concentration. Besides the statistical errors calculated by the fit (Figure 3a) additional systematic errors are the length of the cell and the determination of the HONO concentration (Table 1). We therefore determine the values and the total error of the dif- ferential absorption cross section to be {5'(354.2 nm, nm) = (4.43_+ 0.21) x ]0-19 cm 2, {5'(354.2 nm, nm)= ( ) x 10 - cm 2 and {5' ( nm, nm) = (3.69 +_ 0.19) x 10 -t9 cm Absolute Absorption Cross Sections Figure 3b shows the absolute optical density ln((x)/o(x))/l at three wavelengths as a function of HONO concentration. The data show considerably more scatter than the differential absorption cross sections because of the instability of the lamp intensity, which influences the absolute but not the differential absorption. The problems in assembling the absorption spectrum from the individual spectral intervals confirm this interpretation. Considering the excellent linear behavior of {5'(X1,X2), an error of the HONO concentrations causing the scatter of {5(X) appears improbable. The values of {5(X) were determined by a linear least squares fit weighted with the error of the intensity fluctuation of 0.2%. The errors of the fit results reflect the statistical variability of the optical density. Together with the cell length uncertainty and the error of the determination of the HONO concentration in the cell (Table 1), a total error can be derived. The absorption cross sections and their errors are {5(342 nm)= ( )x cm 2 {5(354 nm) = ( ) x cm 2, and {5(368 nm) = ( ) x 10-1 cm 2. Several other systematic errors also have to be considered. n order to assemble the absorption cross-section spectrum shown in Figure 2, we assumed {5(405 nm) = 0 cm 2. Although this appears to be reasonable and is in agreement with the literature, we can not exclude a small absorption at this wave- length from our measurements. From the slightly negative values at the UV end of our absorption cross-section spectrum, the conclusion can be drawn that {5(405 nm) is larger than assumed here. The error imposed by the uncertainty of the baseline is 7.5 x 10-2 cm 2. The influence of this error is not large for the peak value at 354 nm, but it has a considerable influence on the photolysis rates in the atmosphere as discussed below. The baseline error is an absolute error and has to be distinguished from relative scaling errors described above. A third systematic error is the possibility of steps at the joining wavelengths of the spectral windows [Harder et al., 1997] which is difficult to quantify. Although no steps can be seen above the noise level, they cannot completely be excluded Comparison of the Cross Sections With Literature Data A comparison of our absorption cross-section spectrum with the ones of Stockwell and Calvert [1978] and Bongartz et al. [1991, 1994] is shown in Figure 4. The agreement is good, taking into account the limited spectral quality of the spectrum of Stockwell and Calvert [1978]. Our cross-section spectrum shows a better signal-to-noise ratio than the one by Bongart et al. [1991, 1994], and no residual NO2 absorptions can be observed. Table 3 gives an overview of so far published HONO absorption cross sections. An important aspect when comparing the different absorption cross sections {5(X) is the spectral resolution at which they were measured [Bongartz et al., 1991; Harder et al. 1997]. Therefore we reduced the resolution of all cross-section spectra by a convolution with a Gaus- sian function of 1 nm full width at half maximum (FWHM). n the case of Johnston and Graham [1974] and Stockwell and Calvert [1978], we repeatedly applied a triangular smoothing procedure until the width of the HONO bands agreed with those of our measurements reduced to 1 nm resolution. n all resolution-reduced absorption cross-section spectra, the widths of the HONO absorption bands were compared to ensure comparable bandwidths.

6 ß 14,590 STUTZ ET AL.' UV-VSBLE CROSS SECTONS OF NTROUS ACD 6xl 0 ' ' 5xl 0 9 4xl 0-9 i i i Stockwell and Calvert E 3x C ' 2x10 ' * Bongartz et al xlO -2ø cm 2 i,, i ,,, ting procedure clearly shows uncorrected NO2 absorptions which increase the absolute c ( ), while c '( ) is not influenced. While the agreement in c ( ) with the values of Stockwell and Calvert [1978] is reasonable, a much higher difference is observed in c '( ). We suspecthat this is caused by the low spectral resolution of their measurements, although we took special efforts to adapt the width of the HONO bands. The value of Perner and Platt [1979] was adapted to our resolution by using the method described by Bongartz et al. [1991, 1994], because the spectrum was not available. t agrees within the uncertainty of this method. To compare the i, i, i 36o 380 4oo value given by Febo et al., [1996], we applied a high-pass wavelength (rim) filter similar to the one applied on this cross-section spectrum. This filter reduces c '( ) considerably, which explains the low Figure 4. Comparison of the absorption cross section spectra value given in this paper. The agreement with the c '( ) specof this work with available literature values. The agreement trum of Vasudev [1990] and Bongartz et al. [1991, 1994] is with the measurements of Bongartz et al. [1991, 1994] is excellent. The cross section of Stockwell and Calvert [1978] excellent. As in the case of the absolute absorption cross secshows the effect of a low spectral resolution. An offset has tions the values by Brust et al. [1999] disagree with our rebeen added to the literature values to make the comparison sults. t is unclear why their differential absorption cross seceasier. tion spectrum is smaller by -27%. The older measurements of Stockwell and Calvert [1978] are in reasonable agreement with our results, while the values of Cox and Derwent [1976] are 26% higher. The latter can be explained by an uncorrected NO2 absorption, which is clearly visible in their cross section. The disagreement with Johnston and Graham [1974] was already explained by Stockwell and Calvert [1978] by a nonequilibrium state of the former authors' NO - NO2 - H20 mixture during the measurements. The agreement of c ( ) at 1 nm resolution (Table 3) of the more recent measurements of Bongartz et al. [1991, 1994], Pagsberg et al. [1997], and Vasudev [1990] is better than + 10%. Only the most recent measurement by Brust et al. [1999] disagrees with our results. The reason for this discrepancy is not clear at the moment. A comparison of the differential absorption cross sections is difficult, since their size depends not only on the spectral resolution but also on the filtering procedure applied to the absolute absorption cross-section spectrum to determine c '( ). We therefore chose an approach different from the ones published so far. nstead of comparing the difference of two absolute absorption cross sections at different wavelengths, we used a method designed to analyze atmospheric DOAS spectra. n this algorithm our absorption cross-section spectrum was fitted to a literature c ( ) spectrum together with an additive polynomial by a combination of a linear and a nonlinear least squares fit [Stutz and Platt, 1996]. The nonlinear fit allows the correction of spectral shifts and squeezes between the two c '( ). The advantage of this method is the fact that an extended wavelength region is compared instead of two individual wavelengths. t also allows a quantification of the similarity of the bandwidth and therefore the spectral resolu- tion of two o'(30. Since this method compares two o'(30 spectra, only the deviation from our value is given in Table 3. Similar to the absolute absorption cross section, the spectrum of Johnston and Graham [1974] is -70% smaller than our value. Surprisingly, the agreement with the o' spectrum of Cox and Derwent [1976] is better than in the case of the absolute absorption cross section. The residual of the fit- Careful analysis of the resolution-corrected absolute and differential absorption cross sections available in literature shows a surprisingly good agreement, even with the older data (with the exception of Johnston and Graham [1974]). The differences can largely be attributed to spectroscopic problems. The fitting procedure also revealed that some of the published absorption cross-section spectra contain residual NO2 absorptions. For example, the c (,) of Bongartz et al. [1991, 1994] has an impurity of % [NO2]/[HONO] in the spectrum. Although the differential NO2 absorption structure is visible in the spectrum, the absolute absorption of NO2 appears to be corrected in their c (,)..o o 0.02 o 0.00 ß ß wavelength [nm] ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, : HONO reference + residual erence + residual 1 O pixel Figure 5. Example of a DOAS measurement of HONO and NO2 in the atmosphere in Milan, taly, on May 13, 1998, at The NO2 and HONO mixing ratios observed were ( ) ppb and ( ) ppb, respectively. The length of the light path was 2.5 km. A spectrograph-detector system similar to the one described above, equipped with a 600 g mm - grating, yielding a resolution of 0.55 nm, was used.

7 STUTZ ET AL.: UV-VSEBLE CROSS SECTONS OF NTROUS ACD 14,591 loo lo i i, i, i this work., Bongartz et al. Stockwell and Calvert 0 4'0 solar zenith angle /, /o /'! /. / //,/"7// Figure 6. Photolytic lifetime of HONO in the atmosphere. The comparison with the values calculated on the basis of the absorption cross-section spectra of Bongartz et al. [1991, 1994] and Stockwell and Calvert [1978] shows how sensitive qrnono is to small changes of the baseline in the cross-section spectrum. 4. Discussion The new measurements of the HONO absorption cross sections have interesting implications for atmospheric chemistry and the measurement of HONO in the open atmosphere mplications for DOAS Measurements Many measurements of HONO with DOAS have been published. The reported concentrations are based on the older cross-section spectra listed in Table 3. The most widely used cross-section spectrum before the publication of Bongartz et al. [1991] (and their corrected value in 1994 [Bongartz et al., 1994]) was that of Stockwell and Calvert [1978]. Table 3 shows that the cs'(,) reported by Stockwell and Calvert [1978] is higher by - 30% compared to more recent values and to the value of this work. Since DOAS measurements depend more on the differential cross section, many of the older HONO measurements [e.g., Appel et al., 1990; Lainreel and Cape, 1996, and references therein], should be corrected to be comparable to today's values. More recent measurements used the cs(x,) of Bongartz et al. [1991, 1994], which is in excellent agreement with our value. Figure 5 shows an example of the analysis of an atmospheric absorption spectrum with our new absorption crosssection spectrum during the Pianuar Padania Produzione di Ozono / Limitation of Oxidant Production field campaign in Milan, taly (B. Alicke and J. Stutz, unpublished data, 1998). Owing to the high HONO mixing ratio of ( ) ppb, the HONO absorption reached an optical density of 0.01 (lower trace in Figure 5). The excellent description of the HONO absorptions can clearly be seen. The residual of the fit is due to uncorrected instrumental structures. A common problem in the analysis of atmospheric DOAS spectra is the possible interference with the much stronger NO2 absorption (middle trace in Figure 5). This can be caused by an imperfect spectral alignment of the NO2 relative to the HONO absorptions. Although this offset is usually smaller than the statistical noise of the data, it introduces a systematic error which can cause problems in the interpretation of the data. The high accuracy of the spectral position of the new cross sections helps to avoid this problem mplications for Atmospheric Chemistry Figure 6 shows the photolytic lifetime qjhono of HONO calculated with our absorption cross-section spectrum as a function of solar zenith angle. The actinic fluxes used to derive qrhono were calculated for a standard atmosphere, cloudless sky, and the albedo of concrete (-0.2), as can be found in a city, with the system for transfer of atmospheric radiation model [Ruggaber et al., 1994]. The lifetime at midlatitudes at noon is around min. The actual lifetime might be considerably longer for cloud-covered sky. Also shown is qrhono for the cs(x,) of Stockwell and Calvert [1978], which is about 15% longer than our qrhono. The qrnono calculated with the cs(),) of Bongartz et al. [1991, 1994] is about 20% shorter than our values. One uncertainty in these values is the small offset that might be present in our cs(x,) due to the lamp instabilities. Since we assumed cs(405 nm) = 0 cm 2, it is possible that the absorption cross-section spectrum is -7.5x10-21cm 2 higher. Although the change in the band maximum due to this uncertainty is only -1.5%, its influence on the photolyric lifetime is much larger. Assuming a cros section cs(x,)+ 7.5x10-2 cm 2, qrhono decreases by 13%, moving closer to the value of Bongartz et al. [1991, 1994]. The uncertainty of the photolysis rates calculated with our absorption cross-section spectrum is therefore +15%. This uncertainty translates directly into the calculated OH production by HONO photolysis. 5. Conclusion The spectra of differential and absolute absorption cross sections of nitrous acid have been measured with a highpurity source of HONO. The constant flow of HONO allowed the use of a high spectral resolution with a high signal-tonoise ratio. The spectral positions of the major HONO bands are in excellent agreement with previous measurements. Once the differences in the spectral resolution and the common NO2 impurities in literature data were considered, the agreement with our data was within 15%. The new cs(x,) improves the accuracy of the HONO determination by the DOAS technique, which requires a high spectral quality of the crosssection spectrum. The photolytic lifetime at noon calculated with our data is approximately min and can be used for the quantification of the OH source strength of HONO photolysis in urban air. A considerable uncertainty of the photolytic lifetime of HONO due to the error in the absolute value of our and literature absorption cross sections of-15% remains. Acknowledgment. This project was supported by the European Union in the framework of FORMONA. References Allegrini,., F.D. Santis, V.D. Palo, A. Febo, C. Perrino, and M. Possanzini, Annular denuder method for sampling reactive gases and aerosols in the atmosphere, Sci. Total Environ., 67, 1-16, Appel, B.R., A.M. Winer, Y. Tokiwa, and H.W. Biermann, Com- parison of atmospheric nitrous acid measurements by annular denuder and differential optical absorption systems, Atmos. Environ.,24A, , 1990.

8 14,592 STUTZ ET AL.: UV-VS1]3LE CROSS SECTONS OF NTROUS ACD Bongartz, A., J. Kames, F. Welter, and U. Schurath, Near-UV ab- Nash, T., Nitrous acid in the atmosphere and laboratory experiments sorption cross sections and trans/cis equilibrium of nitrous acid, J. on its photolysis, Tellus, 26, , Phys. Chem., 95, , Pagsberg, P., E. Bjerbakke, E. Ratajczak, and A. Sillesen, Kinetics of Bongartz, A., J. Kames, and U. Schurath, Experimental determina- the gas phase reaction of OH + NO (+M) -> HONO (+M) and the tion of HONO mass accommodation coefficients using two differ- determination of the UV absorption cross section of HONO, ent techniques, J. Atmos. Chem., 18, , Chem. Phys. Lett., 272, , Brust, A.S., K.H. Becker, J. Kleffmann, and P. Wiesen, UV absorp- Perner, D., and U. Platt, Detection of nitrous acid in the atmosphere tion cross sections of nitrous acid, Atmos. Environ., 34, 11-19, by differential optical absorption, Geophys. Res. Lett., 6, , Cox, R.A., The photolysis of gaseous nitrous acid, J. Photochem., 3, Platt, U., Differential optical absorption spectroscopy (DOAS), , Chem. Anal. Set., 127, 27-83, Cox, R.A., and R.G. Derwent, The ultra-violet absorption spectrum Platt, U., and D. Perner, Measurements of atmospheric trace gases by of gaseous nitrous acid, J. Photochem., 6, 23-34, long path differential UV/visible absorption spectroscopy, in Op- Febo, A. and C. Perrino, Prediction and experimental evidence for tical and Laser Remote Sensing, edited by D.A. Killinger and A. high air concentration of nitrous acid in indoor environment. At- Mooradien, pp , Springer-Verlag, New York, mos. Environ., 25, , Ruggaber, A., R. Dlugi, and T. Nakajima, Modelling radiation quan- Febo A., and C. Perrino, Nitrous acid generation, fate and detection, tities and photolysis frequencies in the troposphere, J. Atmos. in Encyclopaedia of Environmental Analysis and Remediation, Chem., 18, , edited by R. A. Meyers, pp , John Wiley, New York, Simon, P. K. and P. K. Dasgupta, Continuous automated measure ment of gaseous nitrous and nitric acids and particulate nitrite and Febo, A., C. Perrino and M. Cortiello, A denuder technique for the nitrate, Environ. Sci. Technol., 29, , measurement of nitrous acid in urban atmospheres, Atmos. Envi- Stockwell, R.W., and J.G. Calvert, The near ultraviolet absorption ron., 27, , spectrum of gaseous HONO and N203, J. Photochem., 8, Febo, A., C. Perrino, M. Gherardi, and R. Sparapani, Evaluation of a 203, high-purity and high-stability continuous generation system for Stutz, J., and U. Platt, Numerical analysis and estimation of the stanitrous acid, Environ. Sci. Technol., 29, , tistical error of differential optical absorption spectroscopy meas- Febo, A., C. Perrino, and. Allegrini, Measurement of nitrous acid in urements with least-squares methods, Appl. Opt., 35, , Milan, taly, by DOAS and diffusion denuders, Atmos. Environ., , , Svensson, R., E. Ljungstrfm, and O. Lindqvist, Kinetics of the reac- Harder, J.W., J.W. Brault, P.V. Johnston, and G.H. Mount, Tem- tion between nitrogen dioxide and water vapour, Atmos. Environ., perature dependent NO2 cross section at high spectral resolution, 21, , J. Geophys. Res., 102, , Vasudev, R., Absorption spectrum and solar photodissociation of Harris, G.W., W.P.L. Carter, A.M. Winer, J.N. Pitts, U. Platt and D. gaseous nitrous acid in the actinic wavelength region, Geophys. Perner, Observations of nitrous acid in the Los Angeles atmos- Res. Lett., 17, , phere and implications for the predictions of ozone-precursor re- Vecera, Z., and P.K Dasgupta, Measurement of atmospheric nitric lationships, Environ. Sci. Technol., 16, , and nitrous acids with a wet effluent diffusion denuder and low- Harrison, R.M., J.D. Peak, and G.M. Collins, Tropospheric cycle of pressure ion chromatography-postcolumn reaction detection, nitrous acid, J. Geophys. Res., 101, 14,429-14,439, Anal. Chem., 63, , Jenkin, M.., R.A. Cox, and D.J. Williams, Laboratory studies of the kinetics of formation of nitrous acid from the thermal reaction of nitrogen dioxide and water vapour, Atmos. Environ., 22, , P. Bruno, A. Febo, and C. Perrino, CNR, nstituto nquinamento Johnston, H.S., and R. Graham, Photochemistry of NOx and HNOx Atmosferico, Via Salaria KM 29,300, C.P. 10, Monterotondo St., Roma, taly. E. S. Kim and U. Platt, nstitut ftir Umweltphysik, NF 229, D Heidelberg, Germany. J. Stutz, Department of Atmospheric Sciences, 7127 Math Sciences, Box , University of California, Los Angeles, CA (jochen@atmos.ucla.edu) compounds, Can. J. Chem., 52, , Kanda, Y., and M. Taira, Chemiluminescent method for continuous monitoring of nitrous acid in ambient air, Anal. Chem., 62, , King, G.W., and D. Moule, The ultraviolet absorption spectrum of nitrous acid in the vapor state, Can. J. Chem., 40, , Lammel, G., and J.N. Cape, Nitrous acid and nitrite in the atmosphere, Chem. Soc. Rev., 25, , (Received October 25, 1999; revised December 16, 1999; accepted December 28, 1999.)